Text

theses

Emerging end-to-end scientific applications that integrate high-end experiments and instruments with large scale simulations and end-user displays, require complex couplings and data sharing between distributed components involving large data volumes and varying hard (in-time data delivery) and soft (in-transit processing) quality of service (QoS) requirements. As a result, enabling efficient data coupling is a key requirement of such workflows. In this thesis, we try to address this in two levels; in data transport level and in data sharing abstraction level. Firstly, we leverage software-defined networking (SDN) to address issues of data transport service control and resource provisioning to meet varying QoS requirements from multiple coupled workflows sharing the same service medium. Specifically, we present a flexible control and a disciplined resource scheduling approach for data transport services for science networks. Furthermore, we emulate an SDN testbed on top of the FutureGrid virtualized testbed and use it to evaluate our approach for a realistic scientific workflow. Our results show that SDN-based control and resource scheduling based on simple intuitive service models can meet the coupling requirements with high resource utilization. Secondly, we present design and implementation of an asynchronous data sharing framework for application couplings over wide-area network. Specifically, presented framework extends shared space abstractions of Dataspaces, which is a data sharing framework for HPC applications, to wide-area scale using a NUMA-like architecture and implementation that leverages advanced data transport technologies like GridFTP and RDMA, and uses predictive prefetching using Markov based models to bring remote data close to the application in-time. Finally, our initial results evaluating the performance of the presented framework show that given some slack time between the data insertion and retrieval queries, latency due to data transport over wide-area network can be efficiently masked for realistic scientific workflows.

ETD_6391

M.S.

Includes bibliographical references

by Mehmet Fatih Aktas

doi:10.7282/T33F4RHV

ETD graduate

The author owns the copyright to this work.